The search for new antithrombotic mechanisms and therapies that may spare hemostasis

Thrombotic cardiovascular diseases, including myocardial infarction and stroke, are the leading cause of death in developed countries.¹  Currently used antithrombotic drugs, including antiplatelet agents and anticoagulants, are associated with significant bleeding risk, which increases morbidity and mortality.^2-4  Must we accept that the inhibition of thrombosis is inextricably linked to compromised hemostasis? In fact, there is growing evidence that the molecular and cellular mechanisms governing hemostasis and thrombosis⁵  can be distinguished or at least separated sufficiently to allow for selective therapeutic interventions. If the pathologic event, thrombosis, can be suppressed with limited perturbation of hemostasis, an antithrombotic strategy with reduced bleeding might be realized. In this article, we summarize recent findings that specific molecular events can suppress thrombosis while sparing hemostasis. Some candidate targets with such discriminatory activities remain platelet-associated or coagulation system targets, but they might still necessitate delicately balancing their antithrombotic and antihemostatic activities. Large-scale trials in different clinical settings will be required to determine whether such agents offer therapeutic advantages over current antithrombotic agents. Nevertheless, the discriminating activity of such agents supports the possibility that thrombosis and hemostasis can be distinguishable events. We focused our attention on the unexpected role of a particular membrane protein on leukocytes that can be targeted to selectively suppress thrombosis without apparent compromise of hemostasis. Blockade of the interaction of integrin Mac-1 on leukocytes with 1 of its cognate ligands, GPIbα, on platelets might afford a sufficiently wide therapeutic window between thrombosis and hemostasis to serve as a safer therapeutic target.

Recent studies have suggested that specific components of the blood coagulation system may be targeted to limit thrombosis without severely comprising hemostasis (Table 1). Falling into this category are factor XI (FXI) and factor XII (FXII) as targets and their inhibitors as therapeutic agents. As reviewed by Weitz and colleagues,^6,7  FXI and FXII have emerged as targets to prevent thrombosis without severe perturbation of hemostasis. Humans with FXI deficiency rarely bleed spontaneously, and individuals with FXII deficiency do not bleed. In fact, the first patient reported with deficiency of FXII died of pulmonary thromboembolic disease.⁸  Nevertheless, mice deficient in FXI or FXII have reduced thrombosis.^9,10  Targeting the activities of these factors with antibodies, antisense oligomers, or aptamers has been reported to inhibit thrombosis in mouse models without or with only modestly increased bleeding times.⁶  Importantly, an antisense oligonucleotide that specifically reduces FXI levels was effective at preventing postoperative venous thromboembolism in patients undergoing elective primary unilateral total knee arthroplasty; however, bleeding was increased with antisense FXI compared with enoxaparin (8% vs 3%).¹¹  Despite the encouraging preclinical and phase 2 clinical data, because FXI deficiency can be associated with bleeding and because FXII deficiency can be associated with thrombosis, it seems likely that careful dosing, monitoring, and/or development of effective antidotes will be needed for these targets to progress toward acceptance.

Another key area of research that supports the feasibility of targeting thrombosis that does not compromise hemostasis is the platelet in general and its ligand-receptor interactions in particular (Table 1). Included in this category is CD40 ligand (CD40L). CD40L is released from platelets during thrombus formation, is a ligand for GIIb-IIIa,¹²  and stimulates outside-in signaling.¹³  Gas6 and its tyrosine kinase receptors (mer, tyro3, and axl)^14,15  are involved in thrombus formation and stabilization in both murine and human platelets. Ephrins and their eph kinase receptors,¹⁶  ATP-gated P2X1 cation channel,¹⁷  and myeloid-related protein-8/14 (MRP-8/14 or S100A8/A9) and its platelet CD36 receptor¹⁸  are ligand-receptor pairings that act within the platelet-platelet contact zone or synapse after the initial aggregation event and ultimately promote thrombus growth and stability.¹⁹  Importantly, mice with genetic defects in each of these ligand-receptor interactions have impaired thrombus formation with apparently intact hemostasis as determined by bleeding time assays. In addition, certain intracellular ligands, which bind to or alter critical interfaces and interactions of receptors such as Gα13^20,21  or integrin peptides^22,23  can interfere with thrombosis without perturbing hemostasis. These studies support the premise that thrombosis and hemostasis are distinguishable events. However, the complex organization of the thrombus and the heterogeneity of platelets within the thrombus²⁴  make it uncertain whether the limitations of current antiplatelet drugs can be significantly reduced. Of concern, for example, is the observation that monoclonal antibody against CD40L was associated with unexpected thrombotic complications.²⁵ 

Interactions of GPIbα with its binding partners have also been considered as antithrombotic targets. ARC1779, an aptamer that blocks binding of the A1 domain of von Willibrand factor (vWF) to GPIbα, has been tested in humans as a potential treatment for patients with refractory thrombocytopenic purpura and type 2b von Willebrand disease^26,27  as well as for patients undergoing carotid endarterectomy.²⁸  However, targeting GPIb carries the stigma of hemostatic perturbations associated with Bernard-Soulier syndrome. Indeed, ARC1779 was effective in reducing thromboembolic signals detected by transcranial Doppler ultrasonography in patient undergoing carotid endarterectomy but was associated with increased perioperative bleeding and anemia.²⁸ 

Inflammation has been linked to all stages of the development of vulnerable atherosclerotic plaque and its thrombotic complications. Leukocyte count²⁹  and leukocyte activation (as determined by plasma levels of the secreted leukocyte enzyme myeloperoxidase)³⁰  are predictive of future myocardial infarction. Circulating leukocyte-platelet conjugates are also predictive of myocardial infarction³¹  and are a more sensitive marker of in vivo platelet activation than platelet surface P-selectin.³²  A recent study from our group³³  provides an independent line of evidence that physiologic hemostasis and pathologic thrombosis are mechanistically different processes. This study implicates the heterotypic cell-cell interactions between leukocytes and platelets in thrombus formation and suggests that the leukocyte could be an antithrombotic target.

In a series of articles written since 2000, our laboratories have identified the essential importance of leukocyte β₂-integrin Mac-1 (also known as α_Mβ₂, CD11b/CD18) and its counter-receptor, platelet GPIbα, in mediating leukocyte-platelet interactions in vitro and in vivo.^34-36  The α_MI domain of Mac-1 contributes broadly to the recognition of many of its ligands³⁷  and specifically to the binding of GPIbα.³⁴  GPIbα is the largest (135 kDa) subunit of the GPIbα/β/GPIX/GPV complex with ∼25 000 copies found on the surface of platelets.³⁸  The primary (broadly accepted) role of GPIbα in hemostasis is its ability to serve as a receptor for vWF and thrombin.³⁸  Indeed, GPIb has been investigated as an antithrombotic target, but the severity of bleeding in patients with GPIb deficiency (ie, Bernard-Soulier syndrome) has dampened enthusiasm for its therapeutic targeting. Working from the Mac-1 side, we localized a binding site for GPIbα within the α_MI domain segment α_M(P²⁰¹-K²¹⁷) by using a homologous scanning mutagenesis strategy³⁵  that was based on the differential binding of GPIbα to the α_MI and α_LI domains. This strategy was complemented by several independent approaches, including synthetic peptides, site-directed mutagenesis, and gain-in-function analyses. Ultimately, 2 amino acids critical for this interaction (human T²¹³ and R²¹⁶) were identified.³⁵  Antibody targeting of α_M(P²⁰¹-K²¹⁷) blocked α_Mβ₂-dependent adhesion to GPIbα, but not to several other Mac-1 ligands. It also inhibited leukocyte accumulation, cellular proliferation, and neointimal thickening after arterial injury,³⁶  and it broadly regulated the biological response to tissue injury in models of glomerulonephritis³⁹  and experimental autoimmune encephalomyelitis.⁴⁰ 

Because leukocyte-platelet interactions bidirectionally induce signals that amplify proinflammatory and prothrombotic cellular responses,⁴¹  we hypothesized that leukocyte Mac-1 engagement of platelet GPIbα is critical for thrombus formation. We reported recently³³  that mice with Mac-1 deficiency (Mac-1^−/−) or mutation of the Mac-1 binding site (mouse S²¹³A/R²¹⁶A) for GPIbα have delayed thrombosis after carotid artery and cremaster microvascular injury without affecting multiple parameters of hemostasis. Antibody and small molecule interference with the Mac-1:GPIbα interaction inhibited thrombosis.

It has been proposed that after initial aggregation, platelets form a synapse that facilitates signaling by membrane-tethered receptor-ligand pairs and localizing secreted and shed ligands that ultimately promote thrombus growth and stability.¹⁹  Our findings suggest that leukocyte-platelet interactions mediated by Mac-1:GPIbα also seem to function in this synapse. Prior work from our laboratories demonstrated that leukocyte engagement of platelet GPIbα via Mac-1 induces platelet outside-in signaling and platelet activation dependent upon Akt phosphorylation.⁴²  Engagement of platelet GPIbα via Mac-1 induces outside-in Mac-1 signaling that leads to phosphorylation of PKC δ and downregulation of Foxp1 in monocytic cells.³³  Thus, blockade of the initial cell-cell conjugation mediated by Mac-1:GPIbα may prevent bidirectional signaling that amplifies thrombus formation and accounts for the effectiveness of Mac-1:GPIbα inhibition in reducing occlusive thrombus formation in vivo.

Our observations suggest a possible target for therapeutic intervention in cardiovascular and thrombotic diseases. In particular, the specificity of antibodies or small molecules with inhibitory action toward Mac-1:GPIbα suggests that it might be possible to inhibit prothrombotic leukocyte-platelet interactions without affecting other Mac-1 functions or perturbing hemostasis. Bleeding times were not prolonged in mice that were deficient in Mac-1 or that had a mutated GPIbα binding site.³³  Our study identified glucosamine as a prototype of 1 small molecule antagonist of the Mac-1:GPIbα interaction. The failure of other molecules with size and composition similar to that of glucosamine to block α_MI domain:GPIbα interaction augers well for the possibility of detailed structure-activity analyses that could identify more potent small molecule antagonists. Our structural modeling study suggested that several closely related binding modes would allow glucosamine to bind in close to T²¹³/R²¹⁶ and interfere with GPIbα binding. Neither glucosamine nor our antipeptide block α_MI P²⁰¹-K²¹⁷ antibody prolonged tail bleeding times in mice.³³  Other small molecules unrelated to glucosamine were also identified as Mac-1:GPIbα inhibitors. Ultimately, solving the crystal structure of glucosamine or other small molecules identified in our screen or of an anti-M2 bound to the α_MI domain may lead to a new class of antithrombotic therapy.

At a more fundamental level, the results of this study suggest that thrombosis and hemorrhage may be uncoupled at the level of the Mac-1:GPIbα interaction. In particular, the specificity of an antibody or small molecule with an inhibitory action selective for Mac-1:GPIbα might inhibit prothrombotic effects of leukocyte-platelet interactions without affecting other Mac-1 functions. In mice, deficiency of Mac-1 did not interfere with multiple parameters of hemostasis including tail bleeding time, platelet activation and spreading, and plasma coagulation activity (ie, activated partial thromboplastin time and thrombin generation). The identification of a new leukocyte-platelet–dependent pathway of thrombosis that is not required for the arrest of bleeding has significant clinical implications and several limitations. First, the bleeding times in mice are only a crude measure of hemostasis, and it is uncertain whether murine models of thrombosis and hemostasis are predictive of human responses. The proposition that antithrombotic function of inhibitors of the Mac-1:GPIbα interaction may translate to human biology is supported by the recent results of the CANTOS (Canakinumab Anti-inflammatory Thrombosis Outcome Study) trial involving more than 10 000 patients in whom an anti-inflammatory drug (a humanized anti-interleukin-1β [IL-1β] antibody that had no lipid-lowering effects) reduced primary and secondary thrombotic cardiovascular end points, which provides direct evidence for the inflammatory hypothesis of atherothrombotic disease.⁴³  Although the antithrombotic effects of the IL-1β antibody were modest but significant, they support the premise that interference with an inflammatory reaction, leukocyte-platelet conjugation mediated by Mac-1:GPIbα interaction, may be antithrombotic in humans.

However, even if a Mac-1:GPIbα inhibitor of appropriate specificity and potency is identified, the route and timing of administration relative to the thrombotic event and adverse effects associated with the dissipation or discontinuation of the antagonist are challenging issues. Extensive preclinical and clinical testing will be required to determine whether the interaction between leukocyte Mac-1 and platelet GPIbα represents an effective and safe target. In our published articles,^33,35  we describe extensive binding studies that used wild-type and mutant α_MI domains to identify the binding site within Mac-1 for GPIbα. The region within Mac-1 (α_M P²⁰¹-K²¹⁷) responsible for binding to GPIbα is distinct from other Mac-1 ligands. Importantly, the binding of multiple other Mac-1 ligands essential for innate immunity, including ICAM-1, iC3b, CD40L, and fibrinogen, was unaffected by mutation of the 2 amino acids required for GPIbα binding. Nonetheless, because patients with deficiency of β₂-integrins, including LFA-1, Mac-1, and p150,95, suffer from recurrent infections⁴⁴  and because Mac-1–deficient mice have leukocytosis and impaired innate immunity and wound healing,⁴⁵  development of specific inhibitors of Mac-1:GPIbα will require careful surveillance for infectious complications. Finally, we have elected to pursue inhibitors that bind Mac-1 rather than GPIbα to avoid thrombocytopenia that may accompany the binding of platelet receptors, albeit rarely. Nevertheless, the preliminary data do position the Mac-1:GPIbα interaction to be considered a novel and targetable mediator of thrombosis with potentially reduced bleeding risk.

The authors thank Kamila Bledzka for her assistance with the manuscript.

This work was supported by National Institutes of Health, National Heart, Lung, and Blood Institute grants P01HL073311 and R01 HL096062 (E.F.P.) and R37 HL57506 and R01 HL126645 (D.I.S.).

Contribution: E.F.P. and D.I.S. wrote the paper; and Y.W. critically reviewed the paper.

Conflict-of-interest disclosure: E.F.P., Y.W., and D.I.S. are coinventers of technology related to Mac-1:GPIbα assigned to Case Western Reserve University and licensed to Sujana Biotech.

Correspondence: Daniel I. Simon, University Hospitals Cleveland Medical Center, 11100 Euclid Ave, Cleveland, OH 44106; e-mail: daniel.simon@uhhospitals.org.